Commit graph

7 commits

Author SHA1 Message Date
Matt Ellis 68d6df47b9 Fix logic condition in publish script 2018-04-05 14:43:03 -07:00
Matt Ellis bed208ea18 Fix NPM tagging
`npm publish`'s default was to tag the package we published with
`latest` tag. The NPM ecosystem has expected semantics around this
tag (it uses it by default if you don't pass a version).

From the NPM Docs:

> Typically, projects only use the latest tag for stable release
> versions, and use other tags for unstable versions such as
> prereleases.

We were not doing this, but now we will.  We'll have a new tag `dev`
which is the latest build out of CI, and we'll tag builds without a
pre-release tag with "latest".
2018-04-05 14:41:19 -07:00
joeduffy f071329238 Update PyPI publish source 2018-03-09 15:27:16 -08:00
Joe Duffy 98aaf12cdf
Reenable Pylint (#1024)
This change uses virtualenv to insulate us from platform differences
in our building of the Python SDK, and to create an isolated Python 2
environment.  This includes meaning we don't need to worry about the
specific location and behavior of Pylint.  I *think* this will work
no matter whether it's Mac, Ubuntu, ArchLinux, Windows, and so on.

We do install to the --user directory in the install target using
`pip install -e`, however, which enables the machine-wide symlinking
that we need to support various workflows.

This fixes pulumi/pulumi#1007.
2018-03-09 15:11:37 -08:00
joeduffy 4f822b19fd Publish to production PyPI 2018-03-07 11:57:38 -08:00
Matt Ellis d78755765e Set ROOT in publish_packages.sh 2018-03-06 19:36:41 -08:00
Matt Ellis 936abeb118 Publish tgz's in the middle of the build, packages at the end
While it's safe to publish the tgz that we use internally for other
repositories that are on "the link plan" after the build completes, we
shouldn't publish packages to NPM and PyPi at that point. There are
two reasons for doing this:

1. Publishing packages before they are tested, which means we could
end up publishing packages that don't work.

2. NPM prevents publishing the same package more than once, so if we
had to re-run the job (due to tests failing for transient issues), the
publish step will start failing, preventing us from running the tests
at all.
2018-03-06 19:20:47 -08:00